Seasonal effects on performance of a biocontrol agent

Habitat trade-offs in the summer and winter performance of the marginata introduced against the intertidal grass Spartina alterniflora in Willapa Bay, Washington

Fritzi S. Grevstad,1 Robin W. Switzer,1 and Miranda S. Wecker1

Summary Spartina alterniflora is invasive in estuaries of the Pacific coast of North America, as well as in Europe, Asia, Australia, and New Zealand. Willapa Bay, located along the southern coast of Washington state, has the largest infestation of invasive S. alterniflora and is the site of the first biocontrol program against this grass. The recently introduced biocontrol agent, (), has exhibited explosive growth during the summer months, followed by severe declines over the winter. Correlations of quantifiable site characteristics with the growth and decline of 12 released populations reveal the habitat favouring P. marginata. Factors favouring population growth during the summer include high host leaf nitrogen and low spider abundance. Winter survival was greatly improved by the presence of intact dead S. alterniflora culms throughout the winter. Interestingly, sites favouring P. marginata population growth in the summer had the lowest survival over the winter. These correlations and trade-offs suggest possible future strategies for enhancing biocontrol through habitat manipulation.

Keywords: biological control, population growth, Prokelisia marginata, Spartina alterniflora, winter survival.

Introduction during the winter months, with frequent storms and a 2.3 to 3.4 m mean tidal range (Sayce 1988). After two In the three years since its first introduction for biolog- of three initial released populations failed to survive the ical control of Spartina alterniflora in Willapa Bay, winter of 2001–02, and the third population only barely Washington State, the planthopper Prokelisia margi- persisted, 12 additional release sites were selected, nata (Delphacidae) has exhibited explosive population based on their relatively protected locations. By using a growth, demonstrated impacts on the target plant in larger number of release sites, we hoped to find at least field cages, and attained local field densities some sites where P. marginata populations would approaching those known to kill the target weed expand rapidly and persist year to year. Additionally, (Grevstad et al. 2003). However, in spite of these by performing periodic population surveys at these encouraging early signs, the long-term persistence and sites and quantifying habitat characteristics, we sought impact of the agent population has been uncertain, due to identify habitat factors associated with improved P. largely to low overwinter survival. The intertidal envi- marginata performance during both the summer and ronment that Spartina invades is particularly harsh winter months. After one year of following these popu- lations, we have gained important clues as to how to

1 Olympic Natural Resources Center, University of Washington, Forks, give this biocontrol program the best chance of Washington 98331 succeeding. Corresponding author: Fritzi Grevstad, Spartina Biocontrol Program, 2907 Pioneer Road, Long Beach, WA 98631 .

523 Proceedings of the XI International Symposium on Biological Control of Weeds

Invasive Spartina foreign introductions (Grevstad et al. 2003), including a full review by the Technical Advisory Group on Spartina alterniflora, commonly called smooth Biological Control of Weeds. In the past, interstate cordgrass or Spartina, is native and ecologically valued introductions of biocontrol agents have been made on the Atlantic coast of North America, but it is intro- without a formal technical review, including one that duced and a serious environmental threat on the Pacific has been harmful to native plants (Louda and O’Brien coast of North America. S. alterniflora and the closely 2002). related S. anglica and S. townsendii are also invasive in Europe, China, Australia, and New Zealand (Aberle 1993). This perennial grass invades estuarine intertidal Prokelisia marginata life history mudflats, which are normally devoid of emergent vege- Prokelisia marginata is native to the Atlantic and tation, dramatically transforming them into expansive Gulf coasts of North America. It also occurs in Cali- swards of tall dense vegetation. The invasion brings fornia, where it may have been introduced in recent threats to a wide variety of birds, fish, and commer- decades. P. marginata is highly host specific, using cially harvested clams and oysters that rely on the only a small number of closely related Spartina spp. as mudflat habitat. hosts (Grevstad et al 2003). In addition to S. alterni- Willapa Bay, a 23,000 hectare estuary along the flora, it can complete development on S. anglica and S. southern Washington coast, has the most advanced foliosa (native to California and Mexico). It may also infestation of invasive S. alternflora. The plant was be capable of using the European S. maritima and S. accidentally introduced as early as the 1890s during a townsendii, although these were not included in period when it was used as packing material for oysters host range tests. P. marginata weakens the plant by shipped from the Atlantic coast (Frenkle and Kunze ingesting sap from the phloem and also by laying eggs 1984). The plant was slow to spread until the mid 1900s under the leaf surface, causing structural damage and when an apparent increase in seed production launched scarring to the leaf. P. marginata is known to have three the population into a phase of rapid expansion (Sayce generations per year in its native range and in California 1988; Feist and Simenstad 2000). Aerial photos docu- (Denno et al. 1996, Roderick 1987) but so far has ment a 60% increase in Spartina cover throughout the produced no more than two generations in per year in bay between 1994 and 1997 (Reeves 1999). In 2002, an Willapa Bay. Nymphs pass through five instars before estimated 2400 solid hectares of S. alterniflora plus moulting into adults. Overwintering occurs in the 2200 hectares of scattered patches were present in nymphal stages. The majority of nymphs pass the Willapa Bay (Wecker et al. this volume). winter inside leaf curls of senesced plants (thatch). Some can also be found on short green shoots, which Novel aspects of the Spartina biocontrol are sparse in winter. program Materials and Methods Several aspects of the Spartina biocontrol program are unique. First, this is the first use of classical biocon- Releases of approximately 9000 mixed stage P. margi- trol against a grass. A lack of projects targeting grasses nata were made at 12 sites throughout Willapa Bay in (Julien and Griffiths 1998) may reflect the fact that late May and early June of 2002. The sites were specif- weedy grasses often have relatives of economic or ically selected for their perceived winter habitat quality. ecological importance and tend to be risky targets. This We selected sites in which at least some of the senesced is not the case for S. alterniflora in Willapa Bay. As a S. alterniflora culms remained intact over the winter. member of the tribe Chlorideae, S. alterniflora has few Such sites tended to be in the upper tidal zones, in small close relatives in North America and none in coastal backwater sloughs, or otherwise protected from winter areas north of the San Francisco Bay area. Second, the storms and wave action. In unprotected and lower tidal biocontrol program is the first in a marine intertidal zone sites, the Spartina culms typically break off and environment. This environment has created unique drift away or become waterlogged and decompose. challenges for the biological control program as used for releases were reared on S. alterni- described in this paper. Third, the use of a planthopper flora in a greenhouse during the winter and spring of agent is unusual. The only other documented plan- 2002. The parent stock was collected from Willapa thopper agent is concinna (Stål), used against field populations in late fall. In mid-to-late May, the Parthenium hysterophorus (L.) and Ambrosia artemisi- were released into field sites by nestling ifolia (L.) in Australia (McFadyen 1985; Julien and infested rearing plants into a designated 5 × 5 m area of Griffiths 1998). Finally, this project differs from most a much larger sward. Most of the planthoppers moved classical biocontrol projects in that the targeted weed is onto nearby field plants within a few days. invasive in the same country where it is native and the The planthopper populations were surveyed at three biocontrol agent has likewise been transferred between times: (1) in early July, before any new eggs had states rather than between countries. The host specifi- hatched; (2) in late September, after one full generation; city testing was nonetheless as rigorous as that used in and (3) in April of the following spring. A gas-powered

524 Seasonal effects on performance of a biocontrol agent vacuum converted from a hand-held leaf blower the decline in density does not necessarily mean a (see Grevstad et al. 2003) was used to sample P. margi- decline in population size. nata. At each release site, insects were vacuumed from A striking pattern to arise from these results is that the vegetation at 12 sample points in July and sites where P. marginata performed well during the September, and at 24 sampling points in the following summer had lowest survival during the winter (Fig. 2). spring (April). At each sample point an area the size of Five of the six populations attaining greater than the intake tube (0.0346 m2) was thoroughly vacuumed. median density appear to have gone extinct, with the Sample points were evenly spaced in a grid arrange- extant population surviving at a rate of only 0.43%. In ment within 5 m radius of the release centre. The contrast, all of the six populations that attained lower vacuum bags were brought back to the laboratory, than median fall densities persisted through the winter where the numbers of P. marginata nymphs and adults and the average survival rate was 8.4%. from each sample were counted. During the September and April surveys, the number of spiders in each sample was also noted. To 100000 assess the possible influence of variation in plant nitrogen on the P. marginata populations, 20 randomly 10000 selected leaves (2nd from top) were collected from each site in mid-September. The leaves were dried in a 2 drying oven, ground to a fine powder, and analyzed for 1000 per m nitrogen content. In April, we quantified characteristics of the wintering habitat inside eight 0.25 m2 quadrats 100 spaced 2 m apart along two transects bisecting the Prokelisia releases area. In each quadrat, we counted the number 10 of new green shoots, measured the height of the tallest shoot, and assessed the percentage of dead culms from 0 the previous year’s growth that were still intact and in July September April good condition. Figure 1. Densities of Prokelisia marginata at 12 release sites in July, September, and April after approx- Results imately 9000 individuals were released at each site in early June. Summer increase At most sites, population densities increased substantially between the first and the second census 0.20 (Fig. 1). The average population increase was by a factor of 2.84 ± 0.90. Change in density ranged from a 50% decline to a nearly 12-fold increase. The change in 0.15 density is an underestimate of the actual reproduction rate because many insects disperse from the initial 0.10 release area. (In an earlier study, roughly two thirds of

the population was found to disperse beyond the imme- Winter survival diate release area by the end of the first summer 0.05 (Grevstad et al. 2003).) The average population density at the end of the summer was 4270 ± 1570 planthoppers per m2 with a range of 947 to just over 20,000 per m2. 0.00 The one site that attained 20,000 per m2 had nearly four 5000 10000 15000 20000 times the density of the next most populous site. Summer performance (planthopper density in September)

Winter decline Figure 2. Relationship between winter and summer Survival over the winter was low, but better than in performance of Prokelisia marginata at 12 previous years. The average fraction surviving from release sites in Willapa Bay. Winter survival was measured as the ratio of spring to fall P. October 2002 to April 2003 was 0.043 ± 0.019. At five marginata densities. Summer performance was sites, no P. marginata were recovered in April. The measured as the planthopper density attained by highest level of survival at a site was 0.18. At all but one September after release of 9000 individuals at site, the density of P. marginata recovered in the spring each site in early June. was lower than densities measured soon after release in the previous summer. Because some insects dispersed,

525 Proceedings of the XI International Symposium on Biological Control of Weeds

Site influences When only the net result of combined summer popu- lation growth and winter declines is considered, i.e. the We found clear correlations between P. marginata density of P. marginata emerging in the spring, the performance and measurable site characteristics. level of intact thatch was the only factor that signifi- During the summer months, P. marginata performance, cantly influenced P. marginata performance (R2 = 0.62, measured as the density attained by the end of the P = 0.003). summer, was positively correlated with leaf nitrogen content (F = 16.5, P = 0.002, Fig. 3a). Leaf nitrogen content at release sites ranged from a low of 1.09% to a Discussion high 2.09%. Summer performance was also strongly Following analyses of the performance of Prokelisia negatively correlated with spider density (F = 11.8, marginata at 12 new release sites, the initial challenges P = 0.006). Spider densities among sites varied by two imposed by the harsh Willapa Bay environment now orders of magnitude with a range of 22 to 2218 per m2 appear surmountable. The careful selection of sites that in September and a range of 2.4 to 176 per m2 in April. were better protected from wind and wave action, as well The outbreak site mentioned above was the site with the as the use of a larger number of varied release locations, highest leaf nitrogen content. It also had the second provided improved overall performance compared to the lowest spider density. first years releases at only three sites. We now also have During the winter, increased survival was strongly three easily quantified habitat factors—high leaf associated with the presence of intact thatch over the nitrogen, low spider density, and the presence of intact 2 winter (R = 0.62; P = 0.002; Fig. 4a). The level of thatch over the winter—that can be used to select future thatch in the quadrats varied among sites from 0 to release sites for even greater improvement in P. margi- 90%, even though all sites had moderate to high levels nata performance. of thatch in the previous spring when the sites were Our results suggest that P. marginata should ideally chosen. Thus, there is variation from year to year in the be released into sites that have high nitrogen and low condition of thatch at particular locations. There spiders in summer and have thatch that remains intact appears to be a threshold level of Spartina thatch over the winter. But such sites may be hard to come by, needed to support P. marginata through the winter. as none of our 12 sites had that combination. Instead Survival was reasonably high at levels of 70% intact nitrogen was negatively correlated with thatch condi- thatch or above, but was low or zero at lower levels. tion and spiders were positively correlated with thatch Interestingly, during the winter, the relationship condition. High nitrogen plants and low spider abun- with spider density was reversed from that in the dance are often found in lower tidal areas and channel summer (R2 = 0.45; P = 0.017; Fig. 4b). P. marginata banks, where there is greater water flow and better survived better at sites where spider densities were access to nutrients, but where the currents and wave high. The likely explanation is that the same conditions action are likely to break off dead culms during the fall that promote P. marginata survival also promote spider and winter. Also, the taller growth of high nitrogen survival. Predation by spiders does not appear to be a plants makes them more susceptible to breakage during significant mortality factor during the winter. The two the fall and winter. As a result of these correlations, other habitat characteristics measured during the spring populations that had explosive growth during the survey, shoot density and culm height, were not signif- summer, reaching sampled densities of 20,000 per m2, icantly correlated with P. marginata survival (Fig. went extinct or nearly so during the winter. In the end, 4c,d). the presence of intact thatch was the only single factor

20000 20000 A. B. 15000 15000

2 10000 2 10000

5000 5000 Plant hoppers per m Plant hoppers per m

1.0 1.2 1.4 1.6 1.8 2.0 2.2 100 1000 Leaf nitrogen (%) Spiders per m2 Figure 3. Relationship between Prokelisia marginata performance and (A) percent nitrogen content of S. alterniflora leaves and (B) spider density.

526 Seasonal effects on performance of a biocontrol agent

0.20 0.20 A. B.

0.15 0.15

0.10 0.10 Winter survival Winter survival 0.05 0.05

0.00 0.00 0 102030405060708090100 0 50 100 150 200 Percent intact thatch Spiders per m2

0.20 0.20 C. D.

0.15 0.15

0.10 0.10 Winter survival Winter survival 0.05 0.05

0.00 0.00 10 20 30 40 50 60 70 80 20 30 40 50 60 Green shoots per m2 Mean shoot height (cm) Figure 4. Prokelisia marginata winter survival as a function of (A) percentage of Spartina thatch remaining intact, (B) spring spider densities, (C) density of green shoots, and (D) mean tallest shoot height. that adequately predicted P. marginata performance expansive swards in the high marsh. In Willapa Bay, over the full year period. only the tall form of S. alterniflora is found and, in all Newly released populations of P. marginata in but the most protected areas, it breaks off during winter. Willapa Bay seem to be foiled by the spatial separation Given that there are very large expanses of Spartina in of superior winter and summer habitat. However, P. Willapa Bay in areas where P. marginata cannot marginata in its native range has a life history strategy survive the winter and only scattered small areas where adapted to it. In New Jersey saltmarshes, P. marginata it can, it is reasonable to question the potential of P. reproduces in tall, nitrogen rich plants along channel marginata to have widespread impact on the target edges during the summer and then disperses in fall to plant over its full distribution. Perhaps a more likely nearby high marsh Spartina that is more favourable for outcome is that the planthopper will have impacts in winter survival (Denno and Grissell 1979). This some areas but not others. dispersal also allows P. marginata to elude predation The results suggest opportunities for habitat manip- by spiders (Denno and Peterson 2000). Such seasonal ulation and conservation biocontrol practices to migration between upper and lower tidal zones has not enhance the effectiveness of the biocontrol program. been observed in Willapa Bay. Instead any dispersal One possibility is to improve P. marginata population that occurs is not directed toward upper tide zones, and growth or even create outbreaks through fertilization of the majority of the planthoppers remain within a few Spartina plants in the vicinity of releases. This could be metres of the release area at the onset of winter done in sites that had good winter habitat and relatively (Grevstad et al. 2003). low spider densities. Fertilization experiments with P. An important difference between east coast and marginata have been tried on the east coast with mixed invasive west coast Spartina marshes is that, on the east results. Bowdish and Stiling (1998) and Denno et al. coast, there are two forms of S. alterniflora; a tall form (1996) found that fertilizing increased P. marginata that grows in lower tide zones and near channel edges, densities by factors of roughly two and four respec- and a short stiff form, 10–15 cm tall, that grows in tively, while Silvanima and Strong (1991) found initial

527 Proceedings of the XI International Symposium on Biological Control of Weeds increases in abundance that did not persist, and Vince et ation in the dispersal strategies of planthoppers. Ecological al. (1981) found no effect of fertilization. Vince et al. Monographs , 389–408. (1981) noted higher numbers of spiders in fertilized Feist, B.E. and Simenstad, C.A. (2000) Expansion rates and plots that may have suppressed the planthoppers. recruitment frequency of exotic smooth cordgrass, Spartina Another approach to enhancing biocontrol is to move alterniflora (Loisel), colonizing unvegetated littoral flats in Willapa Bay, Washington. Estuaries 23, 267–274. large numbers of planthoppers from the high reproduc- Frenkle, R.E. and Kunze, L.M. (1984) Introduction and spread tion sites when they are abundant in the fall and move of three Spartina species in the Pacific Northwest. Annual them to protected locations to spend the winter. Exper- Meeting of the Association of American Geographers, iments are needed to determine what kind of sheltering Washington, D.C. will provide the best winter survival with the least Grevstad, FS, Strong, D.R., Garcia-Rossi, D., Switzer, R.W. effort. The possibility for doing this on a large scale is and Wecker, M.S. (2003) Biological control of Spartina not prohibitive. The state and federal agencies currently alterniflora in Willapa Bay, Washington using the plan- involved in the Spartina control work have large thopper Prokelisia marginata: agent specificity and early machines capable mowing and transporting Spartina results. Biological Control 27, 32–42. stems in large quantities. Julien, M.H. and Griffiths, M.W. (1998) Biological control of weeds: a world catalogue of agents and their target weeds. CABI Publishing, Wallingford, Oxon, U.K. Acknowledgements Louda, S.V. and O’Brien, C.W. (2002) Unexpected ecological effects of distributing the exotic weevil, Larinus planus (F.), This research was supported by the National Sea Grant for the biological control of Canada thistle. Conservation Program and the United States Fish and Wildlife Biology 16, 717–727. Service. We also thank the Washington Department of McFadyen, R.E. (1985) The biological control programme Natural Resources and the Willapa Wildlife Refuge for against Parthenium hysterophorus in Queensland. In airboat transportation, Joe McHugh for site access, and Proceedings of the VI International Symposium on Biolog- Carol O’Casey for field help during the spring survey. ical Control of Weeds (ed. E.S. Delfosse) pp. 789–796, 19– 25 August 1984, Vancouver, Canada. Agriculture Canada, Ottowa. References Roderick, G.K. (1987) Ecology and evolution of dispersal in Californian populations of a salt marsh insect, Prokelisia Aberle, B.L. (1993) The biology and control of introduced marginata. Ph. D. Dissertation, University of California, Spartina (cordgrass) worldwide and recommendations for Berkeley, United States. its control in Washington. Masters thesis, The Evergreen State University, Olympia, WA, United States. Reeves, B. (1999) Report to the legislature: progress of the Bowdish, T.I. and Stiling, P. (1998) The influence of salt and Spartina and Purple loosestrife eradication and control nitrogen on herbivore abundance: direct and indirect effects. programs. Washington State Department of Agriculture, Oecologia 113, 400–405. Olympia, WA, United States. Denno, R.F. and Grissell, E.E. (1979) The adaptiveness of Sayce, K. (1988) Introduced cordgrass, Spartina alterniflora wing-dimorphism in the salt marsh-inhabiting planthopper, Loisel., in salt marshes and tidelands of Willapa Bay, Wash- Prokelisia marginata (Homoptera: Delphacidae). Ecology ington. Unpublished report, Willapa National Wildlife 60, 221–236. Refuge, Naselle, WA, United States. Denno, R.F. and Peterson, M.A. (2000) Caught between the Silvanima, J.V.C. and Strong, D.R. (1991) Is host-plant quality devil and the deep blue sea, mobile planthoppers elude responsible for the population pulses of salt-marsh planthop- natural enemies and deteriorating host plants. American pers (Homoptera: Delphacidae) in northwestern Florida? Entomologist 46, 95–109. Ecological Entomology 16, 221–232. Denno, R.F., Roderick, G.K., Peterson, M.A., Huberty, A.F., Vince, S.W., Valiela, I. and Teal, J.M. (1981) An experimental Dobel, H.G., Eubanks, M. D., Losey, J.E. and Langellotto, study of the structure of herbivorous insect communities in G.A. (1996) Habitat persistence underlies intraspecific vari- a salt marsh. Ecology 62, 1662–1678.

528